Premium
Ultrathin SiO 2 on Si IV. Intensity measurement in XPS and deduced thickness linearity
Author(s) -
Seah M. P.,
Spencer S. J.
Publication year - 2003
Publication title -
surface and interface analysis
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.52
H-Index - 90
eISSN - 1096-9918
pISSN - 0142-2421
DOI - 10.1002/sia.1565
Subject(s) - overlayer , oxide , intensity (physics) , materials science , linearity , work (physics) , bar (unit) , analytical chemistry (journal) , range (aeronautics) , thermal , silicon , substrate (aquarium) , spectral line , scattering , optics , chemistry , thermodynamics , optoelectronics , physics , composite material , oceanography , chromatography , quantum mechanics , astronomy , meteorology , metallurgy , geology
The procedures for measuring the intensities and for subsequent calculation of the thickness of thermal SiO 2 layers on Si in the range 0.3–8 nm have been evaluated to determine the best measurement protocols. This work is based on earlier work where the measurements for (100) and (111) Si surfaces indicate the need to work at a reference geometry. In the spectra, the Si 2p peaks may be separated clearly into the substrate Si and the overlayer SiO 2 but it is recommended here that, for accuracies better than 1%, the interface oxides are also analysed. The analysis here is for thermal oxides. Oxides grown by other routes may require a modification of this analysis. It is shown that, in evaluating the data to determine the layer thickness, the failure to remove x‐ray satellites or the use of a straight‐line background will both lead to unacceptable errors that may exceed 5%. On the other hand, if a Shirley background is used consistently for both the peak area analysis and for evaluation of the ratio of intensities for bulk SiO 2 and Si, R o , the results should be linear over the above range to within ±0.025 nm. This excellent result includes the non‐linearities arising from elastic scattering effects and the data reduction method. Equations are provided, together with a value of R o for Mg and Al x‐rays, to calculate the oxide thicknesses with the above linearity. In order to determine the oxide thickness accurately, the relevant inelastic mean free paths also must be known. Theoretical evaluations are only accurate to 17.4% and so better values need to be obtained by calibration. This paper provides the infrastructure to do this. Crown Copyright © 2003 Published by John Wiley & Sons, Ltd.